Phytoplankton pigments were analyzed using high-performance liquid chromatography (HPLC). Water samples (500 mL) collected from 5 to 300 m depths were filtered on 25-mm Whatman GF/F filters under a gentle vacuum (< 0.013 MPa). The filter samples were folded, blotted with filter paper, and stored at −80°C. Phytoplankton pigments were extracted with sonication in N, N-dimethylformamide (DMF) according to the protocol of Suzuki et al. (2005). HPLC pigment was analyzed according to the method of Van Heukelem and Thomas (2001), except that the flow rate was 1.2 mL min⁻¹. Details of the analysis method are in Suzuki et al. (2014).

To estimate the contribution of each algal group in the two layers (5–20 and 30–50 m depths) to total Chl a biomass, the pigment data were interpreted by factorization using the CHEMTAX program (Mackey et al., 1996). The initial and final pigment ratios were calculated following the method of Latasa (2007) (Supplementary Table 1). Initial ratios were based on Suzuki et al. (2002, 2005, 2009), who estimated the community composition of phytoplankton in the northwestern Pacific. Prymnesiophytes, pelagophytes, and prasinophytes in our CHEMTAX analysis are synonymous with type 3 haptophytes, type 2 chrysophytes, and type 3 prasinophytes according to Mackey et al. (1996), respectively. Chl a biomass of each algal group was calculated by multiplying in situ Chl a concentrations by the CHEMTAX outputs.

To confirm the development and decline in Chl a concentration from spring to summer, we also examined monthly satellite images of that concentration as observed by a MODIS/Aqua satellite sensor. MODIS/Aqua monthly Chl a data (4-km resolution, OC3M algorithm) for the period February–July 2010 and 2011 were obtained from the Distributed Active Archive Center (DAAC)/Goddard Space Flight Center (GSFC), NASA (http://oceancolor.gsfc.nasa.gov/cgi/l3).

Water samples (500 mL) were taken at 5-m depth to count diatoms and coccolithophores, and these samples were fixed with buffered formalin (pH 7.8, 1% final concentration). An aliquot (3–11 mL) of the sample was filtered onto a Nuclepore membrane (1-μm pore size) filter set in a glass funnel (3-mm in diameter at the base) in a vacuum (0.013–0.026 MPa). The filter membrane was rinsed with Milli-Q water to remove salts and immediately dried for 3 h in an oven at 60°C. To identify and count the algal cells, the entire area of the membrane (~7 mm²) was examined by a scanning electron microscope (SEM, VE-8800, Keyence Corp., Japan) at magnification approximately 2,000 ×. Phytoplankton species were identified according to Fukuyo et al. (1990); Tomas et al. (1997), Young et al. (2003), Scott and Marchant (2005), and Round et al. (2007). In our study, the dominant diatom species were defined such that they contributed > 25% to total diatom abundance.

Duplicate water samples (each 2 mL) from 5 to 300 m depths were preserved with paraformaldehyde (0.2% final concentration) and stored at −80°C until analysis on land. An EPICS flow cytometer (XL ADC system, Beckman Coulter, Inc., USA) equipped with a 15-mW air-cooled argon laser exciting at 488 nm and a standard filter setup were used for enumeration of heterotrophic bacteria. Prior to analysis, samples were thawed and drawn through a 35-μm, nylon-mesh-capped Falcon cell strainer (Becton-Dickinson) to remove larger cells. Cells were stained with the nucleic acid stain SYBR Gold solution (Invitrogen). The stock SYBR Gold stain (10⁴-fold concentrations in the commercial solution) was diluted to 10-fold concentrations with Milli-Q water. Twenty-five microliters of the 10-fold SYBR Gold stain was added to the 225-μL bacteria sub-sample and incubated in the dark at room temperature (25°C) for 30 min prior to analysis. The incubated samples were mixed with 250 μL of Flow-Count fluorophores (Beckman Coulter) and then analyzed. Details of the flow cytometric analysis are given in Suzuki et al. (2005).

The POC concentration was determined only for the 5-m depth. The 300-mL water sample was filtered onto pre-combusted Whatman GF/F filters (25-mm in diameter, 450°C for 5 h) under a gentle vacuum (< 0.013 MPa) and stored at −20°C until analysis. On land, the samples were thawed at room temperature, exposed to HCl fumes to remove inorganic carbon, and completely dried in a vacuum desiccator for > 24 h. The POC concentrations on the filters were determined with an online elemental analyzer (FlashEA1112, Thermo Fisher Scientific, Inc., USA).

Samples of DOC were collected from 12 depths between 5 and 300 m. Inline filter holders (PP-47, Advantec MFS, Inc., USA) and tubes (Tygon R-3603, United States Plastic Corp., USA) were pre-cleaned by soaking in 1 M HCl and then rinsed with Milli-Q water. Before sampling, pre-combusted Whatman GF/F filters (47 mm diameter) were set in the inline filter holders, and the tubes were mounted. The water samples were taken by connecting the tube with the spigot of the Niskin bottle. At the start of sampling, filtrates were well drained to pre-wash the tubing, filter holder and filter. Afterward, triplicate samples were collected in pre-combusted 24-mL screw vials with acid-cleaned PTFE septum caps. The samples were immediately stored at −20°C until analysis on land. The frozen samples were thawed and well mixed. DOC concentrations were determined with a total organic carbon analyzer (TOC-V CSH, Shimadzu Corp., Japan). The accuracy and variance of the measured DOC concentrations were checked by analyzing deep seawater reference material provided from the consensus reference material (CRM) program (Prof. Dennis Hansell, University of Miami).

Daily POC and DOC production were estimated using a simulated in situ incubation technique at 5-m depth, except at Station M3 during the May 2011 field campaign. The samples were dispensed into three 300-mL, acid-cleaned polycarbonate bottles (two light and one dark) and inoculated with a solution of NaH¹³CO₃ (99 atom% ¹³C), equivalent to ~10% of total inorganic carbon (TIC) in the seawater. All bottles were incubated for ~24 h. The incubated POC samples were filtered onto pre-combusted Whatman GF/F filters (25 mm in diameter, 450°C for 5 h) with a gentle vacuum (< 0.013 MPa), and the filtrate was collected in pre-combusted, 500-mL glass bottles (450°C for 5 h). The filter and water samples were stored at −20°C until analysis.

The filter samples of POC production were thawed at room temperature, exposed to HCl fumes to remove inorganic carbon, and then completely dried in a vacuum desiccator for > 24 h. POC concentrations on the filters and ¹³C abundance were determined using a mass spectrometer (DELTA V, Thermo Fisher Scientific, Inc., USA) with online elemental analyzer (FlashEA1112, Thermo Fisher Scientific). Daily primary production was calculated according to Equation (1) (Hama et al., 1983):

where a_is is the ¹³C atom% in an incubated sample, a_ns is the ¹³C atom% in a natural (i.e., non-incubated) sample, a_ic is the ¹³C atom% in inorganic carbon, [C_p] is the concentration of POC in the incubated sample, t is the incubation duration (day), and f is the discrimination factor of ¹³C (i.e., 1.025). In our study, TIC concentrations were determined with a total alkalinity analyzer (ATT-05, Kimoto Electric Co., Ltd., Japan).

DOC production was estimated according to Hama and Yanagi (2001) with a few modifications. The frozen samples were thawed at room temperature (25°C) and were desalinated using an electro dialyzer (Micro Acilyzer S3, ASTOM Corp., Japan) equipped with AC-220-550 cartridge (CMX-SB/AMX-SB, ASTOM Corp., Japan). Conductivity of the water samples decreased from ~53 to 3.0 mS cm⁻¹ within 2–3 h. Before and after desalination, the DOC concentrations were examined. The recovery percentage of those concentrations ranged from 62 to 96%, whereas the conductivity decreased to 6% of the initial value (Supplementary Table 2). The desalinated seawater samples were concentrated to ~5 mL with a rotary evaporator at 45°C using a pre-combusted 500-mL eggplant-shaped flask. HCl was added to the concentrated 5-mL samples to reduce the pH to 2. The low-pH concentrates were purged with N₂ gas for ~9 min to remove dissolved inorganic ¹³C. Thereafter, the samples were further concentrated to ~0.5–1 mL with a rotary evaporator at 45°C using a pre-combusted 10-mL pear-shaped flask. The ¹³C atom% of DOC absorbed onto the Whatman GF/F filters combusted in a muffle furnace (450°C, 5 h) were determined using a mass spectrometer (DELTA V, Thermo Fisher Scientific) with an online elemental analyzer (FlashEA1112, Thermo Fisher Scientific). Daily DOC production was calculated according to Equation (2) (Hama and Yanagi, 2001):

where [C_d] is the DOC concentration (mg m⁻³).

Furthermore, we estimated the ratio (PER) of DOC production/total production by phytoplankton according to

Water samples of TEP were taken from depths 5–300 m. Triplicate water samples of 250 mL were filtered using Whatman 0.4-μm Nuclepore filters, applying a gentle vacuum (< 0.013 MPa). TEPs captured by the filter were stained with Alcian blue solution (Passow and Alldredge, 1995). The stained filters were immediately rinsed with Milli-Q water and then stored at −20°C until analysis on land.

For the standard TEP curve, the method of Claquin et al. (2008) was modified. As reported by Claquin et al. (2008), the standard curve described in Passow and Alldredge (1995) is applicable only for low concentrations of TEP (i.e., calibration standard weight of xanthan gum ranging between 0 and 40 μg). One milligram of xanthan gum was put into a 15-mL centrifugal tube and added to 10 mL of the Alcian blue stain. The tube was shaken well for ~1 h to combine the xanthan gum and Alcian blue. Thereafter, it was centrifuged at 3,200 × g for 20 min, and the supernatant liquid was carefully removed using micro pipets. Ethanol (99.5%) was added to the tube, which was then centrifuged, and the supernatant liquid was removed via the ethanol precipitation method. The ethanol precipitation was repeated at least five times until the solution became transparent. Finally, as much of the ethanol in the tube as possible was removed. To dry the blue-stained xanthan gum, N₂ gas was gently sprayed into the tube. Subsequently, the xanthan gum was completely dried in a desiccator under vacuum for more than 24 h. The blue-stained xanthan gum was dissolved with 6 mL of 80% H₂SO₄ using a volumetric pipette (stock solution of 1 mg xanthan gum 6-mL⁻¹). The stock solution was diluted with 80% H₂SO₄, to make a dilution series with five concentrations (10, 50, 100, 150, and 200 μg xanthan gum 6-mL⁻¹). Absorption at 787 nm was measured with a Shimadzu spectrophotometer (MPS-2400) in 1-cm cuvettes, with reference to Milli-Q water. Four slopes of the TEP standard curve were compared with those of Passow and Alldredge (1995) and Claquin et al. (2008) (Supplementary Table 3). Also we conducted a comparison between the standard curves made with this centrifugation and conventional filtration (Passow and Alldredge, 1995) methods. The comparison revealed that the slope (136) of the centrifugation method was lower than that (207) of the conventional method (Supplementary Figure 1). Hence, all TEP concentrations in this study were corrected by multiplying a factor of 1.52. Passow and Alldredge reported that average absorption of the filter blank with 0.4-μm polycarbonate filters was between 0.07 and 0.09. In our study, the filter blank (average ± standard deviation) was 0.081 ± 0.003 (n = 4).

The natural filter samples were transferred into 20-mL vials. Six milliliters of 80% H₂SO₄ was added to the vials, and the filters were soaked for 2 h. The vials were gently agitated five times over this period. Absorption at 787 nm was measured with the spectrophotometer in 1-cm cuvettes against Milli-Q water as reference. The TEP levels (μg xanthan gum equivalent L⁻¹) were determined according to Passow and Alldredge (1995).

Nutrients were sampled every other day. Duplicate water samples per culture vessel were dispensed into acrylic tubes with screw caps, and the samples were immediately stored at −20°C. The frozen samples were thawed at room temperature and well mixed. The nutrient concentrations (nitrate, phosphate, and silicate) were determined with an auto-analyzer (QuAAtro 2-HR, BLTEC Corp., Japan) as described above. Sampling of cell abundance was done every other day. Ten to fifty milliliters of the water samples were transferred into 15 or 50-mL centrifuge tubes. During days 0–14, T. nordenskioeldii was filtered using a 25-mm Whatman Nuclepore membrane filter of 1.0-μm pore size and was immediately resuspended in the non-concentrated water samples. One point five milliliters of the concentrated sample (days 0–14) or non-concentrated sample (days 16–40) was transferred to a micro-slide glass chamber of thickness 1.1 mm (S109502, Matsunami Glass Ind., Ltd., Japan), and covered with a cover glass. To estimate cell abundance, duplicate samples were observed per culture vessel using an epifluorescence microscope (BZ-9000, Keyence Corp., Japan) equipped with a GFP optical filter (OP-66835 BZ filter GFP, Keyence Corp., Japan). The Chl a-derived fluorescence in T. nordenskioeldii was automatically photographed with 28 photos and 4 × magnification, and the fluorescence was summed using image analysis software (Keyence Corp., Japan). Therefore, cell abundance was estimated by dividing the observed volume (mL) by the total value of fluorescence. The specific daily growth rate (μ, d⁻¹) during the culture experiment was estimated by the method of Guillard (1973):

where N₀ is initial cell abundance (cells mL⁻¹), N_t the abundance after t days, t₀ the initial time (d), and t_t the time after t days.

Chl a concentrations were determined in the mid-exponential phase. Triplicate samples per vessel (each 200 mL) were filtered onto Whatman GF/F filters with a gentle vacuum (< 0.013 MPa). The filters were folded, blotted with filter paper, and stored at −80°C. The pigment analysis is described in Section Phytoplankton Pigments and CHEMTAX Processing.

TEP was sampled once every 4 days between days 0 and 36. Triplicate water samples (1–20 mL) per culture vessel were filtered using Whatman 0.4-μm Nuclepore membrane filters in a gentle vacuum (< 0.013 MPa). Post-treatment used the same method described in Section Transparent Exopolymer Particle (TEP) Analysis.

DOC was sampled once every 4 days between days 0 and 36. A filter funnel (25-mm in diameter, PN 4203, PALL Corp., USA) was pre-cleaned by soaking in 1 M HCl, and then rinsed with Milli-Q water. Pre-combusted Whatman GF/F filters were placed in the filter funnel, which was equipped with a suction vessel. The water sample was filtrated in a gentle vacuum (< 0.013 MPa). At the start of sampling, a few milliliters of the filtrates were drained to prewash the GF/F filter and filter funnel. Then, duplicate samples were collected in pre-combusted 24-mL screw vials with acid-cleaned PTFE septum caps. The samples were immediately stored at −20°C until analysis. The analysis method was described in Section Dissolved Organic Carbon (DOC) Concentration.

Based on TEP or DOC concentration in the culture vessels, cellular TEP productivity [pg xanthan gum equiv. (cell)⁻¹ d⁻¹] and DOC productivity [pg C (cell)⁻¹ d⁻¹] averaged during the exponential growth and stationary phases (initial and final) using two data points of those phases were estimated via

where C₀ is TEP concentration (pg xanthan gum equiv. mL⁻¹) or DOC concentration (pg C mL⁻¹) at the initial time of the exponential growth phase or stationary phase, C_t is concentration of the TEP or DOC at the final time after t days, t₀ is the initial time (d) of the exponential growth phase or stationary phase, t_t is the final time after t days, N₀ is initial cell abundance (cells mL⁻¹) of the exponential growth or stationary phase, and N_t is cell abundance at the final time after t days. Carbon-based TEP productivity was estimated by multiplying TEP productivity by a factor of 0.75 (Engel and Passow, 2001).

Seawater temperatures at 5 m increased from April through June 2010 (Table 1). By contrast, salinity only changed slightly throughout the cruises (32.8–33.3). Relatively high nutrient levels were found in April 2010, which declined in June. The surface mixed layer was above 30 m at stations A2, M1, and M2, whereas it was shallower than 10 m at the other stations.

Chl a concentrations generally decreased from April through June 2010 and from 5 to 300 m depths (Figures 2A–C). Satellite surface Chl a images also indicated that decreases in the algal pigment concentration from April to June in 2010 and 2011 (Supplementary Figure 2). The highest Chl a concentration at 5-m depth was observed at station A1 in April 2010 (8.8 μg L⁻¹), whereas the lowest was at station J2 in June 2010 (0.4 μg L⁻¹) (Table 1). Chl a within the mixed layer significantly decreased from April to June (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April vs. May: p < 0.01; April vs. June: p < 0.01; May vs. June: p < 0.01).

The final ratio matrix in the 5–20 m layer estimated with the CHEMTAX program (Supplementary Table 1B) was within the range of Mackey et al. (1996), except for the alloxanthin:Chl a ratio (0.25), which was slightly larger than that of the maximum (0.23) of Mackey et al. The final ratio matrix in the 30–50 m layer (Supplementary Table 1D) was within the range of Mackey et al. The contributions of each phytoplankton group to Chl a biomass from 5 to 50 m depths are shown in Figure 3. Diatoms dominated in both April 2010 (average ± standard deviation: 94 ± 4%) and May 2011 (80 ± 9%). By contrast, the contributions in June 2010 (35 ± 23%) were relatively small. The contributions of the other phytoplankton groups, such as cryptophytes (20 ± 17%), prasinophytes (19 ± 10%) and prymnesiophytes (9 ± 6%), were larger than those in April 2010 (cryptophytes: 1 ± 2%; prasinophytes: 2 ± 2%; prymnesiophytes: 0.4 ± 0.5%) and May 2011 (cryptophytes: 9 ± 5%; prasinophytes: 3 ± 2%; prymnesiophytes: 3 ± 1%).

Cell abundances of diatoms at 5-m depth were relatively high in April 2010 (335 × 10³ cells L⁻¹ on average) and low in June 2010 (5 ± 4 × 10³ cells L⁻¹) (Table 1). Diatom abundance in May 2011 was highly variable (150 ± 115 × 10³ cells L⁻¹). Using SEM, we identified 34 centric diatom species, 13 pennate diatom species, and 3 coccolithophore species, including a holococolith species (Table 2). The dominant diatom species in April 2010 were Thalassiosiara nordenskioeldii and Chaetoceros spp. In June 2010, Fragilariopsis pseudonana and Neodenticula seminae were predominant in the diatom community. In May 2011, Ch. atlanticus and Chaetoceros spp. dominated the diatom assemblages. Chl a concentrations were significantly correlated with the cell abundance of centric diatoms (Spearman's rank correlation ρ = 0.90, p < 0.005, n = 9), but only weakly correlated with that of pennate diatoms (ρ = 0.45, p = 0.17, n = 9). The cell abundance of coccolithophores was relatively low throughout the surveys (0–3.6 × 10³ cells L⁻¹).

Bacterial abundances in April 2010 and May 2011 were higher in the surface waters than at depth (Figures 2D,F, Table 1). However, in June 2011, little difference was observed from 5 to 300 m (Figure 2E, Table 1). Bacterial abundance within the mixed layer changed only slightly between April 2010 (4.8 ± 0.9 × 10⁵ cells mL⁻¹), June 2010 (3.5 ± 0.9 × 10⁵ cells mL⁻¹), and May 2011 (3.8 ± 1.7 × 10⁵ cells mL⁻¹) (Kruskal-Wallis ANOVA, p = 0.15). Bacteria abundances from 5 to 300 m depths were significantly correlated with those of Chl a concentrations (Spearman's rank correlation, ρ = 0.65, p < 0.001, n = 94). Significant relationships between these were also observed on each cruise (Spearman's rank correlation, April: ρ = 0.75, p < 0.001, n = 16; June: ρ = 0.49, p < 0.005, n = 45; May: ρ = 0.94, p < 0.001, n = 33). Furthermore, there was a significant relationship between bacterial abundance and Chl a within the mixed layer (Spearman's rank correlation, ρ = 0.57, p < 0.01, n = 22). Values of bacteria abundance/Chl a concentration within the mixed layer increased significantly over time, e.g., April 2010 [85 ± 40 cells (pg Chl a)⁻¹], May 2011 [242 ± 68 cells (pg Chl a)⁻¹] and June 2010 [524 ± 70 cells (pg Chl a)⁻¹] (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April vs. May: p < 0.01; April vs. June: p < 0.01; May vs. June: p < 0.01).

The POC concentration at 5-m depth decreased from April through June 2010 (Table 1; April 2010: 454 μg L⁻¹ on average; June 2010: 143 ± 8 μg L⁻¹), and that of May 2011 was 203 ± 37 μg L⁻¹. The ratio of POC/Chl a increased from April through June 2010 (April: 77 on average; June: 237 ± 58), and that in May 2011 was 134 ± 25.

DOC concentrations generally decreased from 5 to 300 m depths (Table 1, Figures 2G–I). The highest concentration (~768 μg L⁻¹) was found at 10 and 20 m depths at station J2 and at 10 m at station J3. The lowest concentration (~540 μg L⁻¹) was observed at 300 m depth at stations M2, J3, and J4. The DOC concentration within the mixed layer in June 2010 (744 ± 24 μg L⁻¹) was significantly higher than that in April 2010 (696 ± 12 μg L⁻¹) and May 2011 (708 ± 12 μg L⁻¹) (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April vs. May: p = 0.99; April vs. June: p < 0.05; May vs. June: p < 0.01).

POC production at 5 m depth was between 25 mg C m⁻³ d⁻¹ at the J1 station and 290 mg C m⁻³ d⁻¹ at the A1 station (Table 1). Average productivity in April and June 2010 and May 2011 decreased from April to June (195 ± 95 mg C m⁻³ d⁻¹ in April 2010, 72 ± 29 mg C m⁻³ d⁻¹ in May 2011, and 33 ± 8 mg C m⁻³ d⁻¹ in June 2010). The average DOC production at 5 m depth in April and June 2010 and May 2011 was 2.9 ± 1.3 mg C m⁻³ d⁻¹, 1.9 ± 0.5 mg C m⁻³ d⁻¹, and 3.2 ± 1.0 mg C m⁻³ d⁻¹, respectively (Table 1). PER ratios were estimated. The average ratios increased slightly from April to June (2.3 ± 1.8% in April 2010, 4.6 ± 0.6% in May 2011, and 5.5 ± 1.5% in June 2010) (Table 1).

Minimum and maximum TEP concentrations in our surveys were observed at station A1, at 300 m (<15 μg xanthan gum equiv. L⁻¹) and 20 m (196 ± 71 μg xanthan gum equiv. L⁻¹) respectively (Figures 2K–L, Table 1). TEP concentrations from 5 to 300 m depths were significantly correlated with those of Chl a concentrations throughout the observation period (Spearman's rank correlation, ρ = 0.65, p < 0.001, n = 98) and on each cruise (Spearman's rank correlation, April: ρ = 0.85, p < 0.001, n = 17; June: ρ = 0.52, p < 0.001, n = 47; May: ρ = 0.78, p < 0.001, n = 34). The TEP concentrations were also significantly correlated with bacteria abundance throughout the observation period (Spearman's rank correlation, ρ = 0.65, p < 0.001, n = 98). Additionally, there were significant relationships between the two on each cruise (Spearman's rank correlation, April: ρ = 0.86, p < 0.001, n = 17; June: ρ = 0.36, p < 0.05, n = 48; May: ρ = 0.58, p < 0.001, n = 33).

Average TEP concentrations within the mixed layer in April and June 2010 and May 2011 were 155 ± 12, 90 ± 32, and 112 ± 38 μg xanthan gum equiv. L⁻¹, respectively. TEP levels within the mixed layer in April 2010 were significantly higher than those in June 2010, whereas those in May 2011 were not significantly different from those in April and June 2010 (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April 2010 vs. June 2010: p < 0.01, April 2010 vs. May 2011: p = 0.07; May vs. June: p = 0.48). TEP concentrations within the mixed layer in April and June 2010 and May 2011 were also significantly higher than those below the mixed layer in April (64 ± 49 μg xanthan gum equiv. L⁻¹) and June 2010 (67 ± 27 μg xanthan gum equiv. L⁻¹), and May 2011 (81 ± 23 μg xanthan gum equiv. L⁻¹) (Wilcoxon rank-sum test, April: p < 0.01, n = 17; May: p < 0.05, n = 34; June: p < 0.05, n = 48). TEP values were significantly correlated not only with Chl a concentrations but also diatom-derived Chl a biomass (Spearman's rank correlation, TEP vs. Chl a: ρ = 0.70, p < 0.001, n = 23; TEP vs. diatom-derived Chl a: ρ = 0.66, p < 0.001, n = 23). However, no significant relationship was found between TEP levels and Chl a biomass of other phytoplankton groups (Spearman's rank correlation, TEP vs. cryptophytes-Chl a: ρ = −0.22, p = 0.31, n = 23; TEP vs. prasinophyte-derived Chl a: ρ = −0.27, p = 0.22, n = 23; TEP vs. prymnesiophyte-derived Chl a: ρ = −0.28, p = 0.19, n = 23). The TEP/Chl a ratio increased significantly from April to June (26 ± 10 in April 2010, 76 ± 24 in May 2011, and 128 ± 31 in June 2010) (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April vs. May: p < 0.01; April vs. June: p < 0.01; May vs. June: p < 0.05). In contrast, values of TEP concentration/bacteria abundance changed little from April and June [0.33 ± 0.06 μg xanthan gum (×10⁶ bacterial cells)⁻¹ in April 2010, 0.34 ± 0.16 μg xanthan gum (×10⁶ bacterial cells)⁻¹ in May 2011, and 0.25 ± 0.07 μg xanthan gum (×10⁶ bacterial cells)⁻¹ in June 2010] (Kruskal-Wallis ANOVA, p < 0.01, Steel-Dwass test, April vs. May: p < 0.01; April vs. June: p < 0.01; May vs. June: p < 0.05).

Cell abundances in vessels 1 and 2 increased from 2 to ~20,000 cells mL⁻¹ (Figure 4A). As estimated from the specific growth rate, the exponential growth phase was between days 0 and 30 in Vessel 1 and days 0 and 28 in Vessel 2. The stationary phase was observed between days 30 and 40 in Vessel 1 and days 28 and 40 in Vessel 2. The average specific growth rate in both vessels during the exponential growth and stationary phases was 0.31 ± 0.14 d⁻¹ and 0.03 ± 0.07 d⁻¹, respectively (Table 3). The average nitrate, silicate and phosphate levels in both vessels are shown in Figure 4B. During the exponential growth phase, these levels decreased with time. In the stationary phase, nitrate depleted (< 0.1 μM), whereas silicate and phosphate concentrations maintained values of 3.1 ± 0.7 μM and 0.3 ± 0.03 μM. Chlorophyll a concentrations in Vessel 1 (day 20) and 2 (day 18) in the mid-exponential growth phase were 6.1 ± 0.1 μg L⁻¹ and 7.1 ± 0.1 μg L⁻¹, respectively. Ratios calculated from the Chl a concentrations and cell abundance sampled at the same time were 14 pg cell⁻¹ in Vessel 1 and 13 pg cell⁻¹ in Vessel 2 (average 13 ± 1 pg cell⁻¹).

The initial (day 0) TEP concentration in Vessel 1 was undetectable, so this concentration was assumed to be 0 μg xanthan gum equiv. L⁻¹. The average TEP concentration in both vessels increased with time (Figure 5). The concentrations ranged from 12 ± 12 μg xanthan gum equiv. L⁻¹ to 1,488 ± 93 μg xanthan gum equiv. L⁻¹. Average DOC concentration also generally increased with time (Figure 5). The concentrations were between 1,148 ± 62 μg C L⁻¹ and 1,647 ± 10 μg C L⁻¹. TEP concentrations in the vessels increased with DOC concentrations (ρ = 0.91, p < 0.001, n = 21).

Cellular TEP production rates during the exponential and stationary phases were 2.7 ± 0.5 pg xanthan gum equiv. [cell]⁻¹ d⁻¹ and 8.1 ± 0.9 pg xanthan gum equiv. [cell]⁻¹ d⁻¹, respectively (Table 3). Cellular DOC productivity was 1.0 ± 0.2 pg C [cell]⁻¹ d⁻¹ and 4.0 ± 2.2 pg C [cell]⁻¹ d⁻¹ during the exponential and stationary phases, respectively. The cellular TEP and DOC productivities during the exponential growth phase were also converted to a Chl a-normalized TEP production rate (0.21 μg xanthan gum [μg Chl a]⁻¹ d⁻¹) and DOC production rate [0.07 μg C (μg Chl a)⁻¹ d⁻¹], using the factor of Chl a concentration per cell. The cellular TEP and DOC productivities were 2.9 and 4.0-fold greater in the stationary than exponential growth phase, respectively. Ratios of carbon-based TEP/DOC productivities were 2.2 ± 0.02 in the exponential growth phase and 2.0 ± 0.9 in the stationary phase.